def pull_ncbigene_labels_and_synonyms():
#File format described here: https://ftp.ncbi.nih.gov/gene/DATA/README
ifname = make_local_name('gene_info.gz', subpath='NCBIGene')
labelname = make_local_name('labels', subpath='NCBIGene')
synname = make_local_name('synonyms', subpath='NCBIGene')
bad_gene_types = set(['biological-region', 'other', 'unknown'])
with gzip.open(ifname,
'r') as inf, open(labelname,
'w') as labelfile, open(synname,
'w') as synfile:
h = inf.readline()
for line in inf:
sline = line.decode('utf-8')
x = sline.strip().split('\t')
gene_id = f'NCBIGene:{x[1]}'
symbol = x[2]
gene_type = x[9]
if gene_type in bad_gene_types:
continue
labelfile.write(f'{gene_id}\t{symbol}\n')
syns = set(x[4].split('|'))
syns.add(symbol)
description = x[8]
syns.add(description)
authoritative_symbol = x[10]
syns.add(authoritative_symbol)
authoritative_full_name = x[11]
syns.add(authoritative_full_name)
others = set(x[13].split('|'))
syns.update(others)
for syn in syns:
synfile.write(f'{gene_id}\t{syn}\n')
def pull_hgnc_labels_and_synonyms(infile):
with open(infile, 'r') as data:
hgnc_json = json.load(data)
lname = make_local_name('labels', subpath='HGNC')
sname = make_local_name('synonyms', subpath='HGNC')
with open(lname, 'w') as lfile, open(sname, 'w') as sfile:
for gene in hgnc_json['response']['docs']:
hgnc_id = gene['hgnc_id']
symbol = gene['symbol']
lfile.write(f'{hgnc_id}\t{symbol}\n')
name = gene['name']
sfile.write(
f'{hgnc_id}\thttp://www.geneontology.org/formats/oboInOwl#hasExactSynonym\t{name}\n'
)
if 'alias_symbol' in gene:
alias_symbols = gene['alias_symbol']
for asym in alias_symbols:
sfile.write(
f'{hgnc_id}\thttp://www.geneontology.org/formats/oboInOwl#hasRelatedSynonym\t{asym}\n'
)
if 'alias_name' in gene:
alias_names = gene['alias_name']
for asym in alias_names:
sfile.write(
f'{hgnc_id}\thttp://www.geneontology.org/formats/oboInOwl#hasRelatedSynonym\t{asym}\n'
)
def pull_umls():
"""Run through MRCONSO.RRF creating label and synonym files for UMLS and SNOMEDCT"""
mrcon = os.path.join('input_data', 'MRCONSO.RRF')
rows = defaultdict(list)
priority = read_umls_priority()
snomed_label_name = make_local_name('labels', subpath='SNOMEDCT')
snomed_syn_name = make_local_name('synonyms', subpath='SNOMEDCT')
with open(mrcon,
'r') as inf, open(snomed_label_name,
'w') as snolabels, open(snomed_syn_name,
'w') as snosyns:
for line in inf:
x = line.strip().split('|')
cui = x[0]
lang = x[1]
#Only keep english terms
if lang != 'ENG':
continue
#only keep unsuppressed rows
suppress = x[16]
if suppress == 'O' or suppress == 'E':
continue
source = x[11]
termtype = x[12]
term = x[14]
#While we're here, if this thing is snomed, lets get it
if source == 'SNOMEDCT_US':
snomed_id = f'SNOMEDCT:{x[15]}'
if termtype == 'PT':
snolabels.write(f'{snomed_id}\t{term}\n')
snosyns.write(
f'{snomed_id}\thttp://www.geneontology.org/formats/oboInOwl#hasExactSynonym\t{term}\n'
)
#UMLS is a collection of sources. They pick one of the names from these sources for a concept,
# and that's based on a priority that they define. Here we get the priority for terms so we
# can get the right one for the label
pkey = (source, termtype, suppress)
try:
pri = priority[pkey]
except:
#print(pkey)
pri = 1000000
rows[cui].append((pri, term, line))
lname = make_local_name('labels', subpath='UMLS')
sname = make_local_name('synonyms', subpath='UMLS')
with open(lname, 'w') as labels, open(sname, 'w') as synonyms:
for cui, crows in rows.items():
crows.sort()
labels.write(f'{UMLS}:{cui}\t{crows[0][1]}\n')
syns = set([crow[1] for crow in crows])
for s in syns:
synonyms.write(
f'{UMLS}:{cui}\thttp://www.geneontology.org/formats/oboInOwl#hasExactSynonym\t{s}\n'
)
def pull_prot(which,refresh):
#swissname = pull_via_ftplib('ftp.uniprot.org','/pub/databases/uniprot/current_release/knowledgebase/complete/',f'uniprot_{which}.fasta.gz',decompress_data=True,outfilename=f'uniprot_{which}.fasta')
if refresh:
swissname = pull_via_urllib('ftp://ftp.uniprot.org/pub/databases/uniprot/current_release/knowledgebase/complete/',f'uniprot_{which}.fasta.gz')
else:
swissname = make_local_name(f'uniprot_{which}.fasta')
swissprot_labels = {}
nlines = 0
maxn = 1000
with open(swissname,'r') as inf:
for line in inf:
nlines += 1
if line.startswith('>'):
#example fasta line:
#>sp|Q6GZX4|001R_FRG3G Putative transcription factor 001R OS=Frog virus 3 (isolate Goorha) OX=654924 GN=FV3-001R PE=4 SV=1
x = line.split('|')
uniprotid = f'UniProtKB:{x[1]}'
name = x[2].split(' OS=')[0]
swissprot_labels[uniprotid] = f'{name} ({which})'
#if nlines > maxn:
# break
print('numlines',nlines)
print('nl',len(swissprot_labels))
swissies = [ (k,) for k in swissprot_labels.keys() ]
print('s',len(swissies))
return swissies, swissprot_labels
def pull_prots(refresh_swiss=False,refresh_trembl=False):
swiss,labels = pull_prot('sprot',refresh_swiss)
fname = make_local_name('labels', subpath='UNIPROTKB')
with open(fname,'w') as synonyms:
for k,v in labels.items():
synonyms.write(f'{k}\t{v}\n')
tremb,tlabels = pull_prot('trembl',refresh_trembl)
for k,v in tlabels.items():
synonyms.write(f'{k}\t{v}\n')
def pull_pubchem_labels():
print('LABEL PUBCHEM')
f_name = 'CID-Title.gz'
cname = pull_via_ftp('ftp.ncbi.nlm.nih.gov','/pubchem/Compound/Extras/', f_name, outfilename=f_name)
fname = make_local_name('labels', subpath='PUBCHEM.COMPOUND')
with open(fname, 'w') as outf, gzip.open(cname,mode='rt',encoding='latin-1') as inf:
for line in inf:
x = line.strip().split('\t')
outf.write(f'PUBCHEM.COMPOUND:{x[0]}\t{x[1]}\n')
def __init__(self):
ifname = make_local_name('mesh.nt', subpath='MESH')
from datetime import datetime as dt
print('loading mesh.nt')
start = dt.now()
self.m = pyoxigraph.MemoryStore()
with open(ifname, 'rb') as inf:
self.m.load(inf, 'application/n-triples')
end = dt.now()
print('loading complete')
print(f'took {end-start}')
def pull_hgnc():
data = pull_via_ftp('ftp.ebi.ac.uk', '/pub/databases/genenames/new/json', 'hgnc_complete_set.json')
hgnc_json = loads(data)
lname = make_local_name('labels', subpath='HGNC')
sname = make_local_name('synonyms', subpath='HGNC')
with open(lname,'w') as lfile, open(sname,'w') as sfile:
for gene in hgnc_json['response']['docs']:
hgnc_id =gene['hgnc_id']
symbol = gene['symbol']
lfile.write(f'{hgnc_id}\t{symbol}\n')
name = gene['name']
sfile.write(f'{hgnc_id}\thttp://www.geneontology.org/formats/oboInOwl#hasExactSynonym\t{name}\n')
if 'alias_symbol' in gene:
alias_symbols = gene['alias_symbol']
for asym in alias_symbols:
sfile.write(f'{hgnc_id}\thttp://www.geneontology.org/formats/oboInOwl#hasRelatedSynonym\t{asym}\n')
if 'alias_name' in gene:
alias_names = gene['alias_name']
for asym in alias_names:
sfile.write(f'{hgnc_id}\thttp://www.geneontology.org/formats/oboInOwl#hasRelatedSynonym\t{asym}\n')
def __init__(self):
ifname = make_local_name('rhea.rdf', subpath='RHEA')
from datetime import datetime as dt
print('loading rhea')
start = dt.now()
self.m = pyoxigraph.MemoryStore()
with open(ifname, 'rb') as inf:
self.m.load(inf, 'application/rdf+xml')
end = dt.now()
print('loading complete')
print(f'took {end-start}')
def pull_pubchem_synonyms():
f_name = 'CID-Synonym-filtered.gz'
sname = pull_via_ftp('ftp.ncbi.nlm.nih.gov', '/pubchem/Compound/Extras/', f_name, outfilename=f_name)
fname = make_local_name('synonyms', subpath='PUBCHEM.COMPOUND')
with open(fname, 'w') as outf, gzip.open(sname,mode='rt',encoding='latin-1') as inf:
for line in inf:
x = line.strip().split('\t')
if x[1].startswith('CHEBI'):
continue
if x[1].startswith('SCHEMBL'):
continue
outf.write(f'PUBCHEM.COMPOUND:{x[0]}\thttp://www.geneontology.org/formats/oboInOwl#hasRelatedSynonym\t{x[1]}\n')
def readlabels(which):
swissname = make_local_name(f'UniProtKB/uniprot_{which}.fasta')
swissprot_labels = {}
with open(swissname,'r') as inf:
for line in inf:
if line.startswith('>'):
#example fasta line:
#>sp|Q6GZX4|001R_FRG3G Putative transcription factor 001R OS=Frog virus 3 (isolate Goorha) OX=654924 GN=FV3-001R PE=4 SV=1
x = line.split('|')
uniprotid = f'UniProtKB:{x[1]}'
name = x[2].split(' OS=')[0]
swissprot_labels[uniprotid] = f'{name} ({which})'
return swissprot_labels
def pull_uber_labels(expected):
uber = UberGraph()
labels = uber.get_all_labels()
ldict = defaultdict(set)
for unit in labels:
iri = unit['iri']
p = iri.split(':')[0]
ldict[p].add((unit['iri'], unit['label']))
for p in ldict:
if p not in ['http', 'ro'] and not p.startswith('t') and not '#' in p:
fname = make_local_name('labels', subpath=p)
with open(fname, 'w') as outf:
for unit in ldict[p]:
outf.write(f'{unit[0]}\t{unit[1]}\n')
def pull_uber_synonyms(expected):
uber = UberGraph()
labels = uber.get_all_synonyms()
ldict = defaultdict(set)
for unit in labels:
iri = unit[0]
p = iri.split(':')[0]
ldict[p].add(unit)
#There are some of the ontologies that we don't get synonyms for. But this makes snakemake unhappy so
# we are going to make some zero-length files for it
for p in expected:
if p not in ['http', 'ro'] and not p.startswith('t') and not '#' in p:
fname = make_local_name('synonyms', subpath=p)
with open(fname, 'w') as outf:
for unit in ldict[p]:
outf.write(f'{unit[0]}\t{unit[1]}\t{unit[2]}\n')
def pull_mesh_labels(self):
s = """ PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>
PREFIX meshv: <http://id.nlm.nih.gov/mesh/vocab#>
PREFIX mesh: <http://id.nlm.nih.gov/mesh/>
SELECT DISTINCT ?term ?label
WHERE { ?term rdfs:label ?label }
ORDER BY ?term
"""
ofname = make_local_name('labels', subpath='MESH')
qres = self.m.query(s)
with open(ofname, 'w', encoding='utf8') as outf:
for row in list(qres):
iterm = str(row['term'])
ilabel = str(row['label'])
meshid = iterm[:-1].split('/')[-1]
label = ilabel.strip().split('"')[1]
outf.write(f'{MESH}:{meshid}\t{label}\n')
def pull_ensembl(complete_file):
f = find_datasets()
cols = set([
"ensembl_gene_id", "ensembl_peptide_id", "description",
"external_gene_name", "external_gene_source", "external_synonym",
"chromosome_name", "source", "gene_biotype", "entrezgene_id",
"zfin_id_id", 'mgi_id', 'rgd_id', 'flybase_gene_id', 'sgd_gene',
'wormbase_gene'
])
for ds in f['Dataset_ID']:
print(ds)
outfile = make_local_name('BioMart.tsv', subpath=f'ENSEMBL/{ds}')
#Really, we should let snakemake handle this, but then we would need to put a list of all the 200+ sets in our
# config, and keep it up to date. Maybe you could have a job that gets the datasets and writes a dataset file,
# but then updates the config? That sounds bogus.
if os.path.exists(outfile):
continue
atts = find_attributes(ds)
existingatts = set(atts['Attribute_ID'].to_list())
attsIcanGet = cols.intersection(existingatts)
df = query(attributes=attsIcanGet, filters={}, dataset=ds)
df.to_csv(outfile, index=False, sep='\t')
with open(complete_file, 'w') as outf:
outf.write(f'Downloaded gene sets for {len(f)} data sets.')
